Case study · Regulated & Compliance-Heavy Workflows

Insurance: Scaling P&C underwriting capacity, with authority

Led and delivered by founder Kannan Durairaj, prior role. Client identity withheld.

Challenge

Mid-market submissions arrived through broker emails and portals, with information spread across proposal forms, property schedules, claims histories and quote requests. Underwriters spent valuable time finding, rekeying and checking information before assessing risk. The insurer needed more capacity and faster broker responses, while keeping experienced underwriters accountable for underwriting decisions.

Approach

Prepare the submission. Preserve human judgement.Agents gathered, classified and extracted submission information, giving underwriters a structured view of the risk. Underwriters retained responsibility for assessment and approval.

Match the tool to the task: Agents handled document interpretation and evidence preparation. Deterministic rules applied defined underwriting guidelines, appetite checks and referral triggers.

Make evidence easy to verify: Extracted information and supporting evidence were linked to source documents, helping underwriters review the submission and investigate gaps or exceptions.

Outcome

Quote turnaround fell from a week to under six hours, less time spent gathering and rekeying information gave underwriters more capacity to assess risk without proportionate headcount growth, and the insurer reported $2 million in mid-market gross written premium growth alongside the new workflow.

Broker experience: Quote turnaround fell from a week to under six hours, helping brokers receive timely responses for their clients.
Underwriting capacity: Less time spent gathering and rekeying information gave underwriters more capacity to assess risk, without proportionate headcount growth.
Commercial value: The insurer reported $2 million in mid-market gross written premium growth alongside the new workflow.

Intake and triage

Before: manual and document-heavy: Submissions and documents sorted manually
After: governed Human-AI workflow: Agents capture submissions, classify documents and flag missing information

Data preparation

Before: manual and document-heavy: Details rekeyed across systems
After: governed Human-AI workflow: A structured submission record links extracted information to source documents

Guidelines and appetite

Before: manual and document-heavy: Manual checks
After: governed Human-AI workflow: Rules-based checks flag appetite issues and referral requirements

Risk review

Before: manual and document-heavy: Underwriters assemble the evidence
After: governed Human-AI workflow: Agents prepare a summary of exposures, gaps and exceptions

Decision and response

Before: manual and document-heavy: Preparation work delays assessment
After: governed Human-AI workflow: Underwriters assess the risk and approve the quote or decline

Stage	Before: manual and document-heavy	After: governed Human-AI workflow
Intake and triage	Submissions and documents sorted manually	Agents capture submissions, classify documents and flag missing information
Data preparation	Details rekeyed across systems	A structured submission record links extracted information to source documents
Guidelines and appetite	Manual checks	Rules-based checks flag appetite issues and referral requirements
Risk review	Underwriters assemble the evidence	Agents prepare a summary of exposures, gaps and exceptions
Decision and response	Preparation work delays assessment	Underwriters assess the risk and approve the quote or decline

Takeaway

Prepare information faster without transferring underwriting judgement to AI. Combining document-processing agents, rules-based checks and human approval gives underwriters more capacity to focus on risk and broker relationships.